Uso de análise multivariada para identificação de zonas de potenciais produtivos agrícolas

• Ano de defesa: 2023
• Autor(a) principal: Melos, Natália Duarte
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Agronomia; UFSM; Programa de Pós-Graduação em Agricultura de Precisão; Colégio Politécnico da UFSM
• Palavras-chave em Português: Sensoriamento remoto; Análise de clusters; Mapas de colheita; Remote sensing; Cluster analysis; Harvest maps; CNPQ::CIENCIAS AGRARIAS::AGRONOMIA
• Link de acesso: http://repositorio.ufsm.br/handle/1/29437

The highlight in the Brazilian economy is agribusiness, specifically grain production, which takes place from north to south of the country. Farmers aim to increase their production efficiency with minimal input usage on the same cultivated area. Precision Agriculture aligns with this objective as an essential tool for detecting problems within the property and aiding in decision-making processes. However, some productivity variations cannot be explained solely by traditional mapping of variables; it is necessary to perform analyses with multiple variables to find indicators that contribute to understanding this scenario. The objective of this study was to define zones of productive potential through multivariate analysis for an agricultural cultivation area using data from various sources. The study was conducted in a commercial soybean field covering 35.40 hectares with a sampling grid of one point every two hectares, generating a database with altitude, clay content, soil fertility for the year 2018, fertilization, corrections, and productivity (basic data). These data, combined with remote sensing data (NDVI, NDRE, NDWI, and Surface Brightness Temperature) for the four subsequent harvests following the sampling, increased the degree of data utilization to intermediate and advanced levels. The mesh file was used to create Voronoi polygons, and the data was tabulated and subjected to cluster analysis. The polygons were grouped according to the dendrograms of each analysis based on the data acquisition degree. The productivity data from the four harvests were used to calculate the historical average productivity. The results formed the maps of zones of potential productivity for each dendrogram of each cut and were compared to identify the stabilization of zone formation. To choose the final map representing the management areas, the maps were subjected to map algebra to find the map of zones of potential productivity. Thus, the map that best represented the arrangement of zones of potential productivity belonged to the intermediate category for the 2nd cut of the dendrogram, comprising five zones of potential productivity ranging from very low to very high productivity. The results were satisfactory, and the objective of the study was achieved. The two methodologies for finding management zones and potential productivity converged to the same arrangement in this study, incorporating not only traditional Precision Agriculture data but also remote sensing data. However, in other analyses, it is possible to include data on variable seed rates, rainfall, electrical conductivity, and other variables.